The chronoloGy and periodizaTion of The Bronze and the early iron age burial ground in domasław, wrocław district, based on radiocarbon dating

the article presents the results of the radiocarbon dating and bayesian analysis of 14c dates of bones from the burial ground in domasław. the bayesian analysis used the relative chronology obtained based on the characteristic features of grave goods and the assigning of individual burials to specific periods of the late bronze age (iii eb – V eb) or the early iron age (hc – lta). a coherent chronological model of the burial ground was accepted after assuming that graves with transitional features, attributable to two subsequent periods, could have been contemporary of graves from one or the other period. the temporal frames of particular periods calculated by the model allow us to improve previously published chronological diagrams of the late bronze age and the early iron age in the region. key words: radiocarbon dating, bayesian analysis, bronze age, domasław Tomasz Goslar


Methodology and results of radiocarbon dating of the bones from Domasław
bone samples from 50 cremation graves found at the site in domasław were selected for radiocarbon dating (the programme was implemented as part of the national science centre's project 'transformations in the burial rite in the bronze age and the early iron age as an expression of socio-cultural changes in south-western poland' -contract no. umo-2014/15/b/hs3/02463 signed on 10.08.2015). the bones were dated using the ams technique (accelerator mass spectrometry) at the poznań radiocarbon laboratory (goslar et al. 2004). as a result of the cremation of the bodies, in none of the bone samples (except the youngest one) has collagen been preserved, the substance most often used in the radiocarbon dating of this material. extraction of collagen from the youngest sample (from grave 10336) was carried out using the standard longin method (1971) with the additional use of naoh to remove humic acids, and ultrafiltration (bronk ramsey et al. 2004; brock et al. 2010) to remove shorter amino acid chains resulting from collagen degradation in bones lying in sediment. the satisfactory quality of the obtained collagen (extraction yield 0.7%) was confirmed by the result of measuring the c/n ratio (3.33), which is within the generally accepted admissibility range (van klinken 1999).
in the case of other bones, their nearly white colour indicated a high burning temperature, exceeding 600°c (i.e. calcination). research of lanting et al. (2001) showed that, during the calcination process, recrystallization occurs of apatite, the grains of which become larger and better packed than in the apatite of bones of a living individual. as a result, apatite becomes more resistant to the exchange of carbon with the environment, thanks to which it can be isolated from the less resistant secondary calcite, crystallized in the bone after it has been deposited in the sediment. therefore, the pretreatment of such 14 c dating materials includes the removal of organic fraction residues (by treatment with naclo 2 ) and the removal of possible secondary calcite (by treatment with ch 3 cooh), and only after these steps is the apatite leached to release the carbon designated for 14 c measurement.
as evidenced by later research (zazzo et al. 2009; 2013), during calcination, carbon exchange with the environment may occur, including co 2 contained in the air and co 2 generated during the combustion of cremation fuel. such processes could influence the result of radiocarbon dating if old bones underwent calcination (if cremation was carried out long after the death of individuals), or if the fuel was wood that was very old at the time of cremation (or hard coal). in the case of the bones from graves in domasław, this problem is unlikely.
the results of 14 c dating are shown in table 1. the samples in table 1 are arranged based on relative dating (gediga 1982; 2019), made on the basis of the characteristic features of grave goods. the symbols of particular periods in the table correspond to the division of the late bronze age and the hallstatt period proposed a long time ago by j. kostrzewski (1948), and cited in numerous subsequent publications (e.g. kostrzewski et al. 1965; dąbrowski 2009; gediga 2013). in the relative dating of graves from domasław (gediga, in this volume), which assigns them to the periods of the system by j. kostrzewski, (iii, iV and V in the late bronze age, hc and hd in the hallstatt period), graves with transitional features attributable to both subsequent periods, i.e. ii/iii eb, iii/iV eb, iV/V eb, V eb/hc and hc/hd were distinguished.

Bayesian analysis of the calibrated 14 C dates of samples from Domasław
bayesian analysis is a great tool for the use of chronological information from the 14 c dating of a sample series. by using additional chronological information independent of the 14 c dates, it allows us to establish a more accurate chronology than what is based solely on the results of radiocarbon dating (bayliss 2015, 677-700). amongst the works of polish researchers in this matter that deserve attention are the studies of the 14 c dates series from catalhöyük in turkey (marciniak et al. 2015, 154-176), bukivna in ukraine (makarowicz and goslar 2018, 40-51) and bronocice in poland (kruk et al. 2018, 1-153).
in this paper, the 14 c dates of individual samples were calibrated according to the intcal13 curve (reimer et al. 2013) using the oxcal v4.2.3 program (bronk ramsey 2009; bronk ramsey and lee 2013). the relative dating results were then used as a priori information in the bayesian analysis of the whole set of dates. it was assumed that the dates of individual samples should fall into successive phases, beginning with the oldest iii eb, and ending with the youngest lta, corresponding to the la tène period.
The chronoloGy and periodizaTion of The Bronze and the early iron age burial ground in domasław, wrocław district, based on radiocarbon dating in the first approach (model a, Fig. 1a) it was assumed that graves with transitional features come from the phases absolutely separating the basic periods (e.g. each grave from phase iV/V eb is younger than each grave from phase iV eb and older than each grave from phase V eb). the results of bayesian modelling in this assumption are presented in Fig. 2. the internal coherence of this model is very weak, which is indicated by the value of the agreement index a (bronk ramsey 1995), amounting to amodel: 7, i.e. well below the 60 recommended by bronk ramsey (1995). the reason for the low amodel value is the poor match of the dates of some samples with the model. the dates with low compliance indicators (Fig. 2) are dates from graves c195, c95 and b144 (phase iV/V eb, a: 44, a: 30 and a: 26 respectively), all older than dates of other graves from this phase, dates from graves e83, 8428, 8375, b85 and 5011 (phase iV eb, a: 38, a: 29, a: 29, a: 17 and a: 37 respectively), as well as dates from graves e110 and e220 (phase iii eb, a: 6, and phase ii/iii eb, a: 10).
the poor consistency of the ten dates with the model does not mean that all of these dates are unrepresentative. For example, the inconsistency of the dates of samples c195, c95, b144, e83, 8428, 8375, b85 and 5011 is due to the fact that the calibrated 14 c dates of the first three samples from this group (iV/V eb phase) are most likely older than the calibrated dates of the remaining five (phase iV eb), while relative dating, with the chronological assumption of model a, suggests the opposite sequence of the dates of these samples. as a result, removing one or the other subgroup of samples from the model would automatically eliminate the incompatibility of the dates of the second subgroup of samples with the model. a similar discrepancy applies to samples e110 (phase iii eb) and e220 (phase ii/iii eb).
the above observations make it necessary to revise the assumption of the chronological model. in the revised model (model b, Fig. 1b), it was assumed that graves with equipment exhibiting features of two consecutive periods can be contemporaneous with graves from one or the other period. this assumption is much more reasonable than that of model a, because there is no doubt that changes in the stylistic features of man-made objects (which, being the equipment of studied graves, are the basis for relative dating) were of a fluid nature and objects with design elements of two subsequent periods could be manufactured at the same time as objects with characteristics of one or the other period. bayesian analysis of model b gives a very good agreement index (amodel: 97), definitely confirming its validity, in opposition to the 'artificially rigid' model a.

The chronological framework of the periodization of the late Bronze Age and the early Iron Age on site in Domasław
bayesian analysis, in addition to clarifying information on the dates of individual samples, allows us to determine the absolute dates of time sections corresponding to the subsequent phases of the model (in this case: periods of relative chronology). the results of dating periods identified in the relative dating of graves from domasław (gediga, in this volume) are presented in table 2.
determined by models the 68% confidence intervals of dates of individual periods, are also shown in Fig. 4. it should be noted that the accuracy of determining the chronological frames is different for different periods. in particular, the two youngest periods (hd and lta) are represented in domasław only by individual samples, hence the timeframes of these periods outlined in table 2 should be treated as approximate. on the other hand, to the hc period, the radiocarbon calibration curve has the hallstatt plateau, meaning that the 14 c dates of materials formed between 750 and 400 bc are almost all the same (see Fig. 6). For this reason, it is always difficult to precisely determine the calibrated dates of 14 c samples. a comparison of the modelling results obtained here with the chronological frames of the j. kostrzewski system published by j. dąbrowski (2009) and reproduced in a later publication by b. gediga (2013) indicates (Fig. 4) that the compliance of the domasław model with the dąbrowski scheme (2009) concerns only the youngest border hd -lta. all earlier periods, in the domasław models, look 100-200 years aged with respect to the dąbrowski scheme, with the aging being the higher the older the periods. it is interesting that in the analyzed part of the past, the difference between the calendar age (cal bp) and radiocarbon age ( 14 c bp) changes in a similar way. this difference (Fig. 5), exceeding 200 years around 1500 bc, decreases to about 100 years around 1000 bc and about 500 bc to around zero. this raises the question whether dąbrowski (2009) in his construction of the chronological scheme did not calculate the radiocarbon years of bp for the bc calendar years by simply subtracting 1950 (the bp is counted to 1950 ad), ignoring the shape of the calibration curve. unfortunately, dąbrowski's publication is of a popular nature and the chronological scheme presented there is difficult to analyze based on the current state of knowledge in a strict manner.
if the chronological scheme published by dąbrowski (2009) essentially presents radiocarbon dates reduced by 1950, its improvement would amount to the calibration of these dates. such a 'corrected' scheme ( Fig. 4) would be well suited to the chronological framework of the graves crustaceans), can cause aging of the 14 c date even by hundreds of years (e.g. olsen et al. 2010; shishlina et al. 2014). Finding the contribution of such food to the diet of the dated individual requires analysis of the composition of stable carbon isotopes (δ 13 c), and above all nitrogen (δ 15 n) in the collagen extracted from the bones (e.g. pospieszny and bełka 2015; goslar et al. 2017). unfortunately, the lack of collagen in the examined bones precludes the possibility of such analysis, both in domasław and in kietrz. thus, in this aspect of the analysis of the chronology of cremation graves from domasław, we are doomed to uncertainty.